Labsco
MCP SERVER

Blender MCP Senpai

by xhiroga

Spot n-gons and topology problems in your Blender scene, and get an AI second opinion on the mesh — without running a separate server.

3D, CAD & Game Engines
Summary
The add-on is the server, which is why there is nothing to configure.

Most Blender bridges ask you to run a separate process and keep it alive alongside Blender; here enabling the extension opens the endpoint, and the client just connects. That also fixes the scope: it works while Blender is open on your machine, and the connection is a local SSE port rather than anything remote.

What it is

A Blender add-on that also speaks MCP. It highlights topology issues directly in the viewport and exposes the scene to ChatGPT, Claude or Gemini for improvement suggestions. The server lives inside the add-on, so there is nothing else to install or keep running.

What you get
  • Automatic n-gon highlighting, so topology issues show up as you model rather than at export time
  • AI comments on the mesh from ChatGPT, Claude or Gemini
  • Zero external MCP server — enabling the add-on is the whole setup
  • An SSE endpoint at http://localhost:13180/sse that any MCP client can attach to
  • Asset Store integration is listed as planned, not implemented
Requirements

Blender 4.2+. Install through Blender itself: Edit → Preferences → Get Extensions → Repositories → Add Remote Repository, using https://xhiroga.github.io/blender-mcp-senpai/extensions/index.json, then search for Blender Senpai, install it and enable it under Add-ons. For Claude, Cline or Roo Code, bridge to the SSE endpoint with `npx -y supergateway --sse http://localhost:13180/sse`; Dive connects to that URL directly. Restart the app after configuring — on Windows, end the process from Task Manager. Licensed GPL-3.0.